<p>Passive terahertz (THz) imaging offers a unique capability to detect and characterise objects on the basis of their radiation emission. In this paper, the potential of passive THz imaging for the detection of persons hidden in a non-metallic cargo is investigated. A radiation model has been developed and explored during experiments to analyse capabilities of passive THz imaging for human detection. Since the density and composition of the cargo can significantly impact the effectiveness of human detection and dense materials can obscure human forms, a novel approach is proposed that takes advantage of deep learning techniques to analyse THz images and accurately detect hidden persons. A concept exploring the sequential approach to human detection has been adopted. The person detection and pose estimation algorithm in a cargo algorithm (PDEC) is proposed and validated. The method involves a single training of two tasks on a data set of annotated THz images. Experimental results demonstrate the effectiveness of our proposed method in accurately detecting the presence of humans in various scenarios, including indoor and outdoor environments. The proposed method shows impressive performance, achieving a mean value of DR of 0.93 and the mean value of FDR at 0.02 for nonoccluded images. In case of occlusion appearance, the mean value of DR is 0.89 while the mean value of FDR is 0.04. Our work contributes to the development of advanced surveillance and security systems that can operate in challenging conditions, such as low light or obscured visibility. According to the best of our knowledge, this is the first known work showing capabilities of passive THz imaging for detection of human presence in cargo.</p>